Graphical Representation of Proximity Measures for Multidimensional Data: Classical and Metric Multidimensional...

Area of Science:

• Computational Biology
• Data Science
• Bioinformatics

Background:

• Analyzing complex datasets with multidimensional attributes requires effective visualization techniques.
• Understanding relationships and similarities within data is crucial for scientific discovery.

Purpose of the Study:

• To present classical and metric multidimensional scaling (MDS) for visualizing data proximity.
• To demonstrate the application of MDS in reducing dimensions for data exploration.

Main Methods:

• Employed classical and metric multidimensional scaling (MDS).
• Utilized proximity matrices, eigenvalues, eigenvectors, and numerical minimization for linear and nonlinear mappings.
• Applied MDS to immunological and sequence similarity of influenza proteins, and spatial reconstruction of airport locations.

Main Results:

• Successfully projected high-dimensional data onto two or three dimensions, preserving metric differences.
• Demonstrated the utility of MDS across diverse datasets, including biological and spatial data.
• Highlighted the practical application of MDS in data analysis and interpretation.

Conclusions:

• Classical and metric MDS are powerful tools for exploring and visualizing complex, multidimensional data.
• These methods offer valuable insights into relationships within biological sequences and spatial data.
• The provided Mathematica programs facilitate the application of these MDS techniques.